Japanese 2003 Rules - Informal Text Version
Robert Jasiek
somewhat more ordinary English:
http://home.snafu.de/jasiek/j2003inf.html
Do you have comments or corrections?
--
robert jasiek
-
Can a ruleset properly be termed "Japanese" if no Japanese Go
Organization has formally adopted it ? For all we know, you may
as well call it "Martian 2003 Rules" because an equal number of
Martian Go Organizations have adopted it, i.e. the number zero.
- regards
- jb
-----------------------------------------------------------------
"A round in the esoteric cosmology of Theosophy, Anthroposophy
and Rosicrucianism is a cosmic cycle or sequence by which an
evolving reincarnating being passes through the various stages
of existence as the Earth, the Solar System or the Cosmos comes
into and passes out of manifestation."
http://en.wikipedia.org/wiki/Round_(Theosophy)
-----------------------------------------------------------------
Robert Jasiek
> Can a ruleset properly be termed "Japanese" if no Japanese Go
> Organization has formally adopted it ? For all we know, you may
> as well call it "Martian 2003 Rules" because an equal number of
> Martian Go Organizations have adopted it, i.e. the number zero.
A Japanese picture is not necessarily one that is acknowledged by some
Japanese painting organization. A Japanese ruleset is not necessarily
one that is acknowledged by some Japanese go association. Robert
Superko is not necessarily the name of a go player.
--
robert jasiek
ian
The term "alive stones" is understandable but rather abnormal in the
Inglish language. I would suggest "living stones"
Robert Jasiek
>The term "alive stones" is understandable but rather abnormal in the
>Inglish language. I would suggest "living stones"
Stylistically you are right. OTOH, I do not want to confuse the reader
by using both "alive" and "living" for the same term in the text or by
using a different word for the analogous term in a related ruleset.
Avoiding "capturable-x" is a different story because these terms in
the semi-formal text really sound too mathematical in an informal
text. I hope everybody is as happy with the substitutes I have
inserted for them.
--
robert jasiek
ian
I see. It's a pity you can't use undead stones, as these are normally
used to describe those in-seki.
Robert Jasiek
>I see. It's a pity you can't use undead stones, as these are normally
>used to describe those in-seki.
Yes? I rarely hear untot or undead.
--
robert jasiek
Jonathan Buss
"Live stones" (no 'a') also works fine. It contrasts nicely with "dead
stones".
Jonathan
Robert Jasiek
<jfb...@uwaterloo.ca> wrote:
>"Live stones" (no 'a') also works fine. It contrasts nicely with "dead
>stones".
Yes, but it is easy to waste 90% of one's time of rules design about
such linguistic fine print. This is not essential for a ruleset.
Essential is consistent usage of terms. If you suggest "live", then
you would also say "A stone is 'live' if..." and we have not really
improved the language. The wisdom of judges is to use nothing but
nouns so if anything it would be "strings are in possession of life".
Stylistically this does not gain anything, either. There is always a
conflict between clarity and good style that cannot be fully dissolved
in a long rules text.
--
robert jasiek
-
> jazze...@hotmail.com (-) wrote:
>> Can a ruleset properly be termed "Japanese" if no Japanese Go
>> Organization has formally adopted it ? For all we know, you may
>> as well call it "Martian 2003 Rules" because an equal number of
>> Martian Go Organizations have adopted it, i.e. the number zero.
Robert Jasiek <jas...@snafu.de> wrote:
> A Japanese picture is not necessarily one that is acknowledged by
> some Japanese painting organization. A Japanese ruleset is not
> necessarily one that is acknowledged by some Japanese go
> association. Robert Superko is not necessarily the name of a go player.
A "picture of Japan" or a "picture done in the likeness of how
Japanese do it" is not the same as a "Japanese picture." Though
Volkswagens originated in Germany, they are no longer "German
Cars" (for import purposes) if manufactured in Brazil or Canada.
You had invited remarks and mine tries to discuss nomenclature
propriety, and issue which you have neglected repeatedly and
seem to be deciding it all by yourself beforehand.
If your analogy was to compare H. Robert Superko, M.D., with
your own posting then you did not supply an "OT" designation for
your Subject: line, since you posted to the rec.games.go newsgroup.
- regards
- jb
--------------------------------------------------------------
COMMON SENSE: An interview with Peter Kingsley
http://www.parabola.org/magazine/current_excerpt.php4
--------------------------------------------------------------
Robert Jasiek
> mine tries to discuss nomenclature
> propriety, and issue which you have neglected repeatedly and
> seem to be deciding it all by yourself beforehand.
"Japanese" in the ruleset title means "modelling what originated from
Japan". You are right that it does not mean "originated from Japan".
By your standard, "Japanese rules" used by amateurs outside Japan are
not Japanese because they neither originated from Japan nor agree to
rulesets that originated from Japan.
As long as some call a ruleset used by amateurs outside Japan
"Japanese rules", it is reasonable enough that my ruleset carries
Japanese in its title. Reason: Mine is closer to the Japanese origin
than what amateurs use, if this is measured in percentage of positions
scored as if scored by a ruleset originated from Japan.
--
robert jasiek
Jonathan Buss
> On Thu, 23 Mar 2006 11:53:13 -0500, Jonathan Buss
> <jfb...@uwaterloo.ca> wrote:
>
>>"Live stones" (no 'a') also works fine. It contrasts nicely with "dead
>>stones".
>
>
> Yes, but it is easy to waste 90% of one's time of rules design about
> such linguistic fine print. This is not essential for a ruleset.
Quite true.
> Essential is consistent usage of terms. If you suggest "live", then
> you would also say "A stone is 'live' if..." and we have not really
> improved the language.
Normally, I would say "a stone is 'alive' if ...". It's just one of
those cases where English is strange.
> The wisdom of judges is to use nothing but
> nouns so if anything it would be "strings are in possession of life".
"strings have life" or "strings possess life", please.
> Stylistically this does not gain anything, either. There is always a
> conflict between clarity and good style that cannot be fully dissolved
> in a long rules text.
I appreciate your efforts at investigating the rules and reporting on
the results, irregardful of your particular choice of words or phrasing.
Please keep up the good work!
Jonathan
Robert Jasiek
<jfb...@uwaterloo.ca> wrote:
>"strings have life" or "strings possess life", please.
Yes, I was kidding a bit. Actually I started with "stones that are
alive" but such a short relative clause I did not consider to be good
style, either.
--
robert jasiek
-
> jazze...@hotmail.com (-) wrote:
>> You had invited remarks and mine tries to discuss nomenclature
>> propriety, an issue which you have neglected repeatedly and
Robert Jasiek <jas...@snafu.de> wrote:
> "Japanese" in the ruleset title means "modelling what originated from
> Japan". You are right that it does not mean "originated from Japan".
I am uncertain that you could arrogate the definition. For example,
would you define what "Japanese freedom" means for the Japanese?
Is "modelling" an approriate term to describe what Japanese intend?
Is it really the case that things Japanese have "originated from Japan"?
Owing to the common Japanese term for "original" and "blunder" do
Japanese strive for originality? Is that something to engender pride?
> By your standard, "Japanese rules" used by amateurs outside Japan
> are not Japanese because they neither originated from Japan nor
> agree to rulesets that originated from Japan.
There is an informal sense for "Japanese rules" whose informality
is not being preserved when you title a webpage with the same phrase.
The criterion for whether a Go Ruleset is Japanese does not stem from
its originality or country of origin but whether some formal Japanese Go
Organization has officially adopted the particular Ruleset as definitive,
or as a guideline. If the NiHon-Kiin were to adopt what were Chinese
Rules or Korean Rules as the national standard, this too would become
a (new) "Japanese Ruleset." Until they do, somebody in Germany who
develops a look-alike ruleset would be producing a "German Ruleset."
> As long as some call a ruleset used by amateurs outside Japan
> "Japanese rules", it is reasonable enough that my ruleset carries
> Japanese in its title. Reason: Mine is closer to the Japanese origin
> than what amateurs use, if this is measured in percentage of positions
> scored as if scored by a ruleset originated from Japan.
I'm not aware that there was any "Japanese origin." Genetic drift
evidence shows some kind of historical flow north from the Philippines,
Formosa, Ryukyu Islands, Okinawa, Kyushu, a flow from the Korean
Penninsula, and a flow from the Siberian continent. Closeness to the
explicit positions is not necessarily a criterion by which the Japanese
determine for themselves what most closely resembles the Japanese
Culture or Japanese Way of doing things. I would suggest that you
take a closer look at Rudolph Steiner, an Austrian who became head
of the German Theosophical Society, 1902-1912, if you wish greater
understanding of what is Japanese (our paradoxical culture climate).
- regards
- jb
-------------------------------------------------------------
Eugenics Flourishes in North Korea:
Marxist egalitarians euthanize disabled
http://www.nationalvanguard.org/story.php?id=8371
-------------------------------------------------------------
Robert Jasiek
> The criterion for whether a Go Ruleset is Japanese does not stem from
> its originality or country of origin but whether some formal Japanese Go
> Organization has officially adopted the particular Ruleset as definitive
This is your definition. I do not share it.
> Until they do, somebody in Germany who
> develops a look-alike ruleset would be producing a "German Ruleset."
Likewise.
--
robert jasiek
Jeff Nowakowski
>
> "Japanese" in the ruleset title means "modelling what originated from
> Japan". You are right that it does not mean "originated from Japan".
I think there is cause for confusion, since in the first paragraph you
mention "Japanese 1949" and "Japanese 1989", which are official
rulesets. My guess is that you would like your Japanese 2003 ruleset to
become official, so that is why you chose this name. I won't argue
against your chosen name, but I do think you should make it clear in the
first paragraph that these rules are your own proposal, and not an
official ruleset.
-Jeff
Robert Jasiek
<jef...@ccs.neu.edu> wrote:
>My guess is that you would like your Japanese 2003 ruleset to
>become official
Rather I would like to see something much more applicable to be used,
be it an Area Scoring ruleset, Ikeda Territory I Rules, or something
close to the New Amateur-Japanese Rules.
>I do think you should make it clear in the first paragraph that these rules are your own proposal, and not an
>official ruleset.
Even before the first parapraph there is a link to the commentary page
that states clearly what the rules are. The semi-formal J2003 page has
a pretty explicit copyright section...
--
robert jasiek
-
> jazze...@hotmail.com (-) wrote:
>> The criterion for whether a Go Ruleset is Japanese does not stem from
>> its originality or country of origin but whether some formal Japanese Go
>> Organization has officially adopted the particular Ruleset as definitive
Robert Jasiek <jas...@snafu.de> wrote:
> This is your definition. I do not share it.
Not all Japanese Go Organizations are nationalistic, so it is Go
rather than nationalism which preceeds the representation of what
is Japanese in the Go community. A Ruleset would need to become
widespread, of course, however it is not in character for a Japanese
Organization to declare officially unless it had become widespread.
>> Until they do, somebody in Germany who
>> develops a look-alike ruleset would be producing a "German Ruleset."
> Likewise.
It is comparitively easy to become a citizen of Germany. You may
find it rather impossible, however, to become a citizen of Japan.
Japan (and Italy) may have been Germany's friends during WW-II,
nevertheless they fought against Germany during WW-I.
------------------------------------------
> Robert Jasiek wrote:
>> "Japanese" in the ruleset title means "modelling what originated from
>> Japan". You are right that it does not mean "originated from Japan".
"Jeff Nowakowski" <jef...@ccs.neu.edu> wrote:
> I think there is cause for confusion, since in the first paragraph you
> mention "Japanese 1949" and "Japanese 1989", which are official
> rulesets. My guess is that you would like your Japanese 2003 ruleset
> to become official, so that is why you chose this name. I won't argue
> against your chosen name, but I do think you should make it clear in
> the first paragraph that these rules are your own proposal, and not
> an official ruleset.
Mr. Jasiek seeks to make a name for himself, by slipping in the
authorship for what he proposes could have been "J2003" (already
getting stale for three years). He pitched the Ruleset during some
International Ruleset Conference in Japan but did not obtain any
bites or nibbles on his hook. Making a name for oneself in this
manner is extremely unlikely to occur and may even impede the
possibility of Japanese Organizations from ever adopting this.
As I had previously stated on October 13 and October 14, 2005,
perhaps the "J" in J2003 refers to "Jasiek" rather than to "Japanese."
- regards
- jb
------------------------------------------------------------------
Bush Hires Red China to Protect Border
Communists will be paid to scan for nuclear material
http://news.yahoo.com/s/ap/20060323/ap_on_go_pr_wh/port_security_bahamas
------------------------------------------------------------------
Robert Jasiek
> A Ruleset would need to become
> widespread, of course, however it is not in character for a Japanese
> Organization to declare officially unless it had become widespread.
You should know better.
OT
> It is comparitively easy to become a citizen of Germany.
I would fail now in Hessen, not knowing what Caspar David Friedrichs
painted:)
/OT
--
robert jasiek
-
> jazze...@hotmail.com (-) wrote:
>> A Ruleset would need to become
>> widespread, of course, however it is not in character for a Japanese
>> Organization to declare officially unless it had become widespread.
Robert Jasiek <jas...@snafu.de> wrote:
> You should know better.
How something becomes "widespread" is also at issue: whether
this occurs via (political) promotion or whether it occurs naturally and
informally, by word of mouth, over the course of many decades. One
interviews members of clubs throughout Japan and asks them to offer
descriptions of their Ruleset(s). A professional is commissioned to
present findings to the academic committee. Some adjustment could
be rendered for "ease of understandability and comprehension."
>> It is comparitively easy to become a citizen of Germany.
> I would fail now in Hessen, not knowing what Caspar David Friedrichs
> painted:)
By revealing such test criteria do you make a citizen task easier? The
State Department gets involved whenever authorizing foreign pro visits.
- regards
- jb
-----------------------------------------------------------------
Terrorist Attacks Planned in Europe Linked to 'American'
http://www.nationalvanguard.org/story.php?id=8375
-----------------------------------------------------------------
Makrai Jozsef
Not specific to the new text, but I wonder if some bent-4s are
problematic to J2003:
.---------------------------------.
| # # O # . O # # # . O . # O # # | komi 33
| # O O # . O O O O O O . # O O # | no prisoners
| # O . # # # # # # # # # # . O # |
| . O # # . . . . . . . . # # O . | W to move
`---------------------------------'
J89: If W passes, B can also pass and wins (all W dead in confirmation,
with pass-for-ko). So W starts a ko and wins it with the central
threat, then gains a further point by starting and losing the other ko
as well. (=Jigo)
Your rules: W corners are alive (in confirmation, W can start both kos
to win one after a (global) ko pass - right?), so W passes. If B passes
he loses. So B approaches, then W starts and wins the *other* ko with
the central threat. Then W make the extra point like above (by forcing
B to win the other ko and/or capture). (=W+1)
B loses on the forced approach move which ends up in his territory, if
the above is correct - is it?
József
dnc AT fw DOT hu
Robert Jasiek
wrote:
>Not specific to the new text, but I wonder if some bent-4s are
>problematic to J2003:
>
>.---------------------------------.
>| # # O # . O # # # . O . # O # # | komi 33
>| # O O # . O O O O O O . # O O # | no prisoners
>| # O . # # # # # # # # # # . O # |
>| . O # # . . . . . . . . # # O . | W to move
>`---------------------------------'
>
>to win one after a (global) ko pass - right?), so W passes. If B passes
>he loses. So B approaches, then W starts and wins the *other* ko with
>the central threat. Then W make the extra point like above (by forcing
>B to win the other ko and/or capture). (=W+1)
J2003: Each string is analysed extra. So each W string is neither
uncapturable, nor capturable-1. This makes the whole board the local-2
of each W string.
Is, e.g., the left W string capturable-2?
l = play on left side
m = play in middle
r = play on right side
#[lllllrrrrPlrPrpp]
etc.
So, yes, the left W string is capturable-2.
Is this problematic to J2003? I don't think so. It does not agree to
J1989's judgement on this position, right; but this is not a reason
because J1989 failed in some positions due to the J1989 hypothetical
ko rule. The relevant criterion is what Japanese professionals would
have thought about this position.
They have never studied (and published) positions with 2 bent-4s. So
we cannot really know. But this your position or n-tuple kos (n>=5)
may well be the positions that pose the greatest doubts about J2003
from our suspected view of Japanese professionals. However,
considering that under the basic-ko rule the behaviour is yet
different from both hypothetical ko rules, I'd rather say that the
professionals would simply be confused themselves. When I asked a
Korean professional 7p about ko threats, rules, and a position with
two identical ko shapes, he felt insulted how anybody could ever ask
such a question that exhibits his own limits of understanding. Hence
presumably every pro would give a different opinion on your position.
Since J2003's philosophy is to give some score rather than output
UNDEFINED in such cases, I am not really worried by your position.
J2003 may have some freedom of choice in positions where professionals
can be expected to use some freedom of choice, too.
--
robert jasiek
Makrai Jozsef
>> .---------------------------------.
>> | # # O # . O # # # . O . # O # # | komi 33
>> | # O O # . O O O O O O . # O O # | no prisoners
>> | # O . # # # # # # # # # # . O # |
>> | . O # # . . . . . . . . # # O . | W to move
>> `---------------------------------'
>>
>> Your rules: W corners are alive (in confirmation, W can start both kos
>> to win one after a (global) ko pass - right?), so W passes. If B passes
>> he loses. So B approaches, then W starts and wins the *other* ko with
>> the central threat. Then W make the extra point like above (by forcing
>> B to win the other ko and/or capture). (=W+1)
>
> J2003: Each string is analysed extra. So each W string is neither
> uncapturable, nor capturable-1. This makes the whole board the local-2
> of each W string.
>
> Is, e.g., the left W string capturable-2?
> l = play on left side
> m = play in middle
> r = play on right side
>
> #[lllllrrrrPlrPrpp]
> etc.
>
> So, yes, the left W string is capturable-2.
>
> Is this problematic to J2003? I don't think so.
Ok, I always said every ruleset is fine if it satisfies the author. ;o)
> It does not agree to
> J1989's judgement on this position, right; but this is not a reason
> because J1989 failed in some positions due to the J1989 hypothetical
> ko rule. The relevant criterion is what Japanese professionals would
> have thought about this position.
What J2003 seems to disagree here is not J89 in particular, but the
idea of Japanese L/D with ko locality...
Using the global ko-pass rule in a Japanese-style ruleset is risky,
since it is a global move that affects all kos together. Your rules
rely on enabling-locality to cover this inherent difference, but such
patching cannot be absolutely safe.
IMO, a more robust rules approach is to keep theoretical concepts
separated and directly correspond to their relevant rule(s), without a
rule relying on side effects of another, irrelevant rule. That way,
should a problem be found, the background concept can be examined to
see if it is at fault. If not (which is usually the case), the
corresponding rule(s) can be improved/refined to reflect the concept
better - without worrying about breaking some other concept or
irrelevant rule.
This approach also makes the rules modular: the concepts can be freely
added/changed/removed, forming a ruleset with any desired features. So,
keeping the concepts of LIFE (with global enabling) and KO (with
optional locality) separated would be advantageous.
Robert Jasiek
wrote:
>What J2003 seems to disagree here is not J89 in particular, but the
>idea of Japanese L/D with ko locality...
This is somewhat of a myth. LJRG and J2003 are the first
Japanese-style ruleset with explicit local environments. J1949 and
WAGC do not have that except implicitly hidden in some particular
precendents. The only explicit local thing in J1989 about ko (apart
from the definition of the ko shape) is the hypothetical ko rule. This
was an artificial invention. In retrospect, one might conclude that ko
during LD analysis is local _because_ J1989 defines it as local. Not
very convincing.
>Using the global ko-pass rule in a Japanese-style ruleset is risky,
Winfried Borchardt has suggested a multi-phase analysis: First analyse
with soft ko rules to determine the local-x of each string, second
analyse with strictly local ko rules to determine the alive-x status
of each string. Or something like that. It has not been tested yet for
many examples. IOW, any ko ruleset is risky because it depends also on
move-sequences and not just on the used ko rules what a status is.
>since it is a global move that affects all kos together.
If the local environment is the whole board, likewise. You can exclude
other kos from every local environment. However, it is much tougher to
exclude all other intersections on that a ko might (re)appear later.
If you want to be really local, then you have to forbid all plays as
ko threats and allow only passes and ko-passes as ko threats;) This is
nonsense, of course, because it lets all ko stones be dead.
Ko fights are global by nature. To make them local, you have to
restrict plays to a subboard. Then there is little choice than to use
something like Winfried's idea. For that, to determine local ko
environments, you have to start with play in global environments. Not
really that much local what you would create as local...
A desire for local arose from the perception of local
two-eye-formations that partition the board into their local parts.
However, when there are such local parts, there is no ko any longer!
Cause and consequence are confused.
>This approach also makes the rules modular: the concepts can be freely
>added/changed/removed, forming a ruleset with any desired features. So,
>keeping the concepts of LIFE (with global enabling) and KO (with
>optional locality) separated would be advantageous.
Rules are never really modular. Ko rules apply to moves that are parts
of sequences that contain moves throughout the whole board or apply to
moves that are parts of sequences that follow a sequence that contains
moves throughout the whole board.
--
robert jasiek
Makrai Jozsef
>> Using the global ko-pass rule in a Japanese-style ruleset is risky,
>
> Winfried Borchardt has suggested a multi-phase analysis: First analyse
> with soft ko rules to determine the local-x of each string, second
> analyse with strictly local ko rules to determine the alive-x status
> of each string. Or something like that. It has not been tested yet for
> many examples. IOW, any ko ruleset is risky because it depends also on
> move-sequences and not just on the used ko rules what a status is.
I meant there's a particular risk about global ko-pass, since it is
KNOWN FROM THE START to contain an undesired feature. It forbids
certain kinds of ko interaction - while allowing others. There is
always the danger that this intrinsic feature or flaw will "leak out"
and become visible. Consider a modified version of this last example:
.---------------------------------.
| # # O # . O # # # . O . # O # # |
| # O O # . O O O O O O . # O O # |
| # O . # # # # # # # # # # . O # |
| . O # # . . . . . . . . # # O . |
`---------------------------------'
.---------------------------------.
| # # O # . O # # # . O . # . . . |
| # O O # . O O O O O O . # . . . |
| # O . # # # # # # # # # # . . . |
| . O # # . . . . . . . . # # . . |
`---------------------------------'
With normal ko, W is alive in both cases (can live even if opponent
moves first). Similar if the game continues. Under pass-for-ko (no ko
fights allowed), W is dead in both cases (can't live even if he moves
first). Both rules has some logic/explanation behind their ruling - one
way or another.
By your rules, W is alive in the first exaple and dead in the second.
But it is hard to accept this just as a random consequence of a random
rule. Such behaviour may need some kind of conceptual explanation. Why
would you want to grant W life in the first example, and not in the
second one?
(And what if there is an uncapturable W group somewhere between the
corners?)
>> the idea of Japanese L/D with ko locality...
>
> This is somewhat of a myth. LJRG and J2003 are the first
> Japanese-style ruleset with explicit local environments. J1949 and
> WAGC do not have that except implicitly hidden in some particular
> precendents. The only explicit local thing in J1989 about ko (apart
> from the definition of the ko shape) is the hypothetical ko rule. This
> was an artificial invention. In retrospect, one might conclude that ko
> during LD analysis is local _because_ J1989 defines it as local. Not
> very convincing.
AFAIK J89 preserved most earlier rulings with rare differences (3pts
w\o capturing comes to mind, but is not ko related). So I think J89's
definition is not just an artificial invention, it (double ko flaw
aside) captures a great amount of truth and knowledge about Japanese
rules.
Explicit local environments: IMO there is another weakness of the
global ko-pass rule to be observed here. For localization, from:
- ko restriction based locality (J89)
- explicit local environments (LJRG)
I think it is important to choose one or the other - but not both.
Mixing the two means trying to do the same job twice, which is an
unfavorable design method.
> You can exclude
> other kos from every local environment. However, it is much tougher to
> exclude all other intersections on that a ko might (re)appear later.
I'm not sure about messing with reappearing kos. Without special
handling they revert to the normal ko rule, and the results will likely
be natural. First capture in a newly formed ko is never restricted, for
example.
> A desire for local arose from the perception of local
> two-eye-formations that partition the board into their local parts.
I think the (Japanese) desire for local may be related to the fact that
it is easier to define "life" without ko fights.
> Rules are never really modular. Ko rules apply to moves that are parts
> of sequences that contain moves throughout the whole board or apply to
> moves that are parts of sequences that follow a sequence that contains
> moves throughout the whole board.
I don't see problems with the whole board. The game is global,
strategic "life" is global too. The players are familiar with global
sequences. Even J89 confirmation assumes global play.
Imagine a generic L/D based non-localizing territory ruleset (my
favorite but not an easy one). There is a noticeable amount of
information and knowledge that is shared between this ruleset and
Japanese-style rules. Stones are alive if there is no profitable attack
on them (in an imaginary continuation). It would be nice to grab this
concept without saying anything about the ko rule used in the
continuation. Then this could be used with any ko rules desired, normal
ko, pass-for-ko (or an improvement) for Japanese-style, or maybe even
superko. (But J89's "life" or similar is NOT yet sufficient, it fails
miserably in some ko fights, has hidden dependencies on pass-for-ko.)
Besides aesthetic and theoretical reasons, aiming for modularity also
has a practical advantage: it would make it easier for (TS) rules to
evolve. So there would be better chances that in time, many unnecessary
concepts and complications could drop out, like seki exception, suicide
exception, etc. - and finally, if the life rule is up to the task,
maybe localization itself. ;o)
Robert Jasiek
wrote:
>there's a particular risk about global ko-pass, since it is
>KNOWN FROM THE START to contain an undesired feature.
I'd say: "...to contain a global feature."
Whether "global" is undesired is then another matter.
If you want local play related to a ko, then also J1989 is not good
enough because it allows global play related to a ko-pass for a
particular ko. E.g., the board might be partitioned by opposing U/C1
strings that are not basic ko stones and then further partitioned by
one's own U/C1 strings that are not basic ko stones. Then for some
local environment hypothetical ko rules could require play only
within.
>.---------------------------------.
>| # # O # . O # # # . O . # O # # |
>| # O O # . O O O O O O . # O O # |
>| # O . # # # # # # # # # # . O # |
>| . O # # . . . . . . . . # # O . |
>`---------------------------------'
>.---------------------------------.
>| # # O # . O # # # . O . # . . . |
>| # O O # . O O O O O O . # . . . |
>| # O . # # # # # # # # # # . . . |
>| . O # # . . . . . . . . # # . . |
>`---------------------------------'
>
>With normal ko, W is alive in both cases (can live even if opponent
>moves first). Similar if the game continues. Under pass-for-ko (no ko
>fights allowed), W is dead in both cases (can't live even if he moves
>first). Both rules has some logic/explanation behind their ruling - one
>way or another.
>
>By your rules, W is alive in the first exaple and dead in the second.
>But it is hard to accept this just as a random consequence of a random
>rule. Such behaviour may need some kind of conceptual explanation. Why
>would you want to grant W life in the first example, and not in the
>second one?
The consequence is not random and the rule is not random. The
consequence is deterministic. The rule is freely chosen within some
possible range given by the set of all previously known, desired
outcomes. Why life / death here? Because in the first example it does
not matter because J pro tradition has not judged about it and because
in the second example J pro tradition has judged dead for the scoring
(but not during the alternation).
>AFAIK J89 preserved most earlier rulings with rare differences (3pts
>w\o capturing comes to mind, but is not ko related). So I think J89's
>definition is not just an artificial invention
Indeed; the rather elegant L2/C2 would not work in general if J89 had
been just artificial.
>- ko restriction based locality (J89)
>- explicit local environments (LJRG)
>
>I think it is important to choose one or the other - but not both.
>Mixing the two means trying to do the same job twice, which is an
>unfavorable design method.
This is your personal opinion but not objective.
>I think the (Japanese) desire for local may be related to the fact that
>it is easier to define "life" without ko fights.
This may also be a reason.
>I don't see problems with the whole board. The game is global,
Sakai insists: "Local is beautiful!" ;) (He refers this to scoring.)
>Imagine a generic L/D based non-localizing territory ruleset (my
>favorite but not an easy one).
What do you mean by generic here?
>Stones are alive if there is no profitable attack
>on them (in an imaginary continuation).
Profitable to mean what? Is this a consequence of your (desired) rules
and you get a vicious circle?
>Besides aesthetic and theoretical reasons, aiming for modularity also
>has a practical advantage: it would make it easier for (TS) rules to
>evolve. So there would be better chances that in time, many unnecessary
>concepts and complications could drop out
I have already created many rulesets that are simplifications of
L/D-TS. All share one thing: Seki is not created due to (!)
exceptional rules.
For rules, seki is an artificial construct. Seki ought to occur as a
strategic object only, even with L/D-TS.
--
robert jasiek
Makrai Jozsef
>> Imagine a generic L/D based non-localizing territory ruleset (my
>> favorite but not an easy one).
>
> What do you mean by generic here?
Without any unnecessary changes or pecularities like seki exception,
suicide exception etc. Best approximation of the reference area game.
>> Stones are alive if there is no profitable attack
>> on them (in an imaginary continuation).
>
> Profitable to mean what? Is this a consequence of your (desired) rules
> and you get a vicious circle?
I don't think so. I meant in a broad human strategic sense. Finding a
strict definition is the hard part, OC.
>> I don't see problems with the whole board. The game is global,
>
> Sakai insists: "Local is beautiful!" ;) (He refers this to scoring.)
But what is "local", afterall? ;o) Explicit local regions and J89's ko
localization are quite different things, for example. Some time ago I
created a similar counterexample for an early "life" idea of mine (with
normal ko):
.-----------------------------------.
| O O # . # . # . # # # # . . . # O |
| O # # # O # # # O . . . O O O # O |
| O # O O O O O O O . . O . . . # O |
| . # O . . . . . . . O . . . . # . | + big W threat(s) elsewhere
| # # O . . . . . . . . . . . . # # |
| O O O O O O O O O O O O O O O O O |
| . . . . . . . . . . . . . . . . . |
`-----------------------------------'
Suppose W's threat would take several moves to remove (eg. something
like B at right). B is dead: W can approach and force the start of the
ko, which W would win. But doing this he would waste several moves
(points). So W does nothing. B could start to remove W's threat(s)
elsewhere and force W to take action in time, but it would cost him
even more moves. So B does nothing too. Strategically, the game is over
and needs statuses and the score.
The problem is that the global ko balance (that caused the strategic
stop) may be distorted during LD analysis. If we only look at this part
of the board, for example, the status changes. So W may still be forced
to perform in the game what is, practically, an explicit and costly
capture from inside his territory. (IMO a 100% clean capture: W never
need to enable anything or ignore a threat.)
Since this also seemed challenging for LJRG, I also discussed it with
Robert Pauli. (Under LJRG, W adds separator stones.) He concluded that
this is an unavoidable consequence of (explicit) localization. I think
he may be right, but this may cast some doubts on localization in
general.
J89's pass-for-ko seems to work well here, but just a coincidence
(global ko balance happens to match the "first to take a ko wins it"
idea). This is often the case, but the opposite also seems possible:
.-------------------------.
| O . # O O O O O O O O O |
| O # # O . . . . . . . . | + big B threat(s) elsewhere, that
| O # O O # # # # # # # # | would take several moves to remove
| . # O . # . # . # . # . |
`-------------------------'
Now W looks dead: B can approach outside and capture all (but lose
points in the process). W can start to remove Bs threat (which doesn't
help him, and also loses points). Nobody needs or wants to play, but
the rules may force B and rob him from his points. (Again a clean
capture, B won't enable anything or ignore a threat).
Specific rulesets aside, how do you think these kind of positions
should be interpreted under territory scoring? Is it acceptable to
force such costly captures from inside territory? (I'm also interested
in how others feel about this)
Playing them out with pass stones, for example, produces results that
are close to the human strategic idea and feel natural. But if pass
stones (dame fill, points on one-sided dame, etc.), why not simply area
scoring? And if area scoring, why not play a more refined game without
actually filling dame, with a good rule for score approximation based
on, say, territory + prisoners? ;o)
Robert Jasiek
wrote:
>>> Imagine a generic L/D based non-localizing territory ruleset (my
>>> favorite but not an easy one).
>> What do you mean by generic here?
>Without any unnecessary changes or pecularities like seki exception,
>suicide exception etc. Best approximation of the reference area game.
"A stone is _alive_ if it is on the board at the end of the
alternation. A stone is _dead_ if it was removed during the
alternation."
Very easy what you call "not an easy one" :)
>> Profitable
>Finding a strict definition is the hard part, OC.
Only if you also demand a quick determination by an algorithm.
>But what is "local", afterall? ;o)
Sakai has not protested to my definitions ;)
>He concluded that [a change in the ko threat situation] is an
>unavoidable consequence of (explicit) localization.
Not really, but an escape would be very artificial: Make all prior ko
threats abstract and keep that abstract environment of threats. A la
Berlekamp/Spight, you know.
>I think
>he may be right, but this may cast some doubts on localization in
>general.
Surprise;)
>Specific rulesets aside, how do you think these kind of positions
>should be interpreted under territory scoring?
We? We are free to choose because nobody before has really thought it
through. Even if Borchardt will feel sick. (He hates such freedom.)
>why not simply area scoring?
How often are you going to ask to convince yourself? ;)
--
robert jasiek
Makrai Jozsef
>>>> Imagine a generic L/D based non-localizing territory ruleset (my
>>>> favorite but not an easy one).
>>>
>>> What do you mean by generic here?
>>
>> Without any unnecessary changes or pecularities like seki exception,
>> suicide exception etc. Best approximation of the reference area game.
>
> "A stone is _alive_ if it is on the board at the end of the
> alternation. A stone is _dead_ if it was removed during the
> alternation."
>
> Very easy what you call "not an easy one" :)
Ok. But is this truly the LD rule for territory scoring, which
gives the best approximation of the reference area score? ;o)
Robert Jasiek
wrote:
>But is this truly the LD rule for territory scoring, which
>gives the best approximation of the reference area score? ;o)
Yes because it describes the nature of area scoring to score all what
is on the board at the end and to score nothing that was removed
before.
--
robert jasiek
ian
Robert Jasiek wrote:
> Due to popular demand I have translated the Japanese 2003 Rules into
> somewhat more ordinary English:
>
> http://home.snafu.de/jasiek/j2003inf.html
>
> Do you have comments or corrections?
>
Handicap Placement is not specified.
Robert Jasiek
>Handicap Placement is not specified.
Yes. It is not the purpose of that ruleset to include any "tournament
rules".
--
robert jasiek
ian
Perhaps. However most tournament rulesets presume that 'Japanese Rules'
dictate that handicap be placed in specific locations. This is why I
raise the point. The EGF rules contain a careless lack of information
on this subject. They merely state that placement of the stones counts
as the first move. I hope this will be updated too.